IN JULY OF 2009, THE HOUSE DEMOCRATS PROPOSED A

Transcription

1 WINDOWS INTO PUBLIC ATTITUDES TOWARDS REDISTRIBUTION Rebbecca Reed-Arthurs, University of California, Davis Steven M. Sheffrin, Tulane University INTRODUCTION IN JULY OF 2009, THE HOUSE DEMOCRATS PROPOSED A new income tax surcharge to finance a major reform of the U.S. health care system. These taxes were geared exclusively to upper-income taxpayers, the top 1.2 percent. The reactions of politicians, tax experts, and commentators to the proposed surtax provide a window into the diverse and often conflicting attitudes that the public holds on income redistribution through taxation. In this paper, we initially look at the reactions to the proposed tax and then present research shedding further light on public attitudes towards redistribution. The proposed surcharge for couples in the House bill called for an additional 1 percent of income between $350,000 and $500,000, 1.5 percent for incomes between $500,000 and $1,000,000, and 5.4 percent on incomes in excess of $1 million with lower thresholds for single filers. Moreover, if savings in health costs envisioned by the bill did not materialize, the bottom two surcharge rates would increase to 2 and 3 percent, respectively. Assuming, as under current law, that the top federal tax rate increases to 39.6 percent and including state income taxes, the Tax Foundation calculated that marginal effective tax rates for the highest brackets would exceed 50 percent in 39 states (Tax Foundation, 2009). One of the first reactions came from Democrat Congressmen who represented high-income districts. After the last election, 14 of the top 25 richest districts were represented by Democrats. Several Democrats voted against the bill and others journeyed to the White House to express their concerns to the chief of staff. While it is one thing to campaign for taxing the rich, the game changes when abstractions on taxing the rich turn into reality. 1 When even the 100 th most expensive zip code has a median house price of $1.62 million, some affected taxpayers with $350,000 family income would not consider themselves as especially wealthy. 2 Prior research suggests that people with high incomes are less likely to support redistribution, but that attitudes are not homogenous among income groups. 3 Congressmen, even from wealthy districts, are thus likely to face some divergence of opinions. This leads us to our first research question: 1. How diverse are attitudes towards redistribution within and between the very low-income and the very high-income taxpayers? A second reaction, from tax policy experts, was that taxing the rich to pay for universal health care was poor policy. There were many variants of this objection. Several were political in nature; for example, the rich should be taxed to finance our existing structural deficits, not for new programs, or that the political equilibrium of taxing the rich to pay for mass programs was unstable. 4 But another critique of the policy stressed the benefit view of taxation, which is rooted deeply in psychological notions of equity theory and social exchange theory. The public expects some linkages between taxes paid and benefits received. Joseph Thor ndike of Tax Analysts made this argument explicit: Financing a mass program with a class tax is not a good idea.it obscures the connection between taxes paid and taxes received a connection that s necessarily tenuous for many government programs, but not for health care. (Bartlett, 2009) Our second research question is: 2. How important are perceptions of benefits of government programs with respect to support for taxation and redistribution? We explore both of these topics in the sections below. DATA Recent surveys have shown some support for taxation and redistribution similar to that proposed in the health care reform package. This paper will focus on one such survey, the Tax Foundation s 2009 Survey of U.S. Attitudes on Taxes, Government Spending and Wealth Distribution, (Moon, 2009). This survey asked 2,002 adults 165

2 NATIONAL TAX ASSOCIATION PROCEEDINGS detailed questions about the fairness of the current tax system and the government s role in income redistribution. Of particular interest to our analysis was the following question: Would you support or oppose the government redistributing wealth by a much higher income tax on high income earners? Respondents rated their reaction as strongly oppose, somewhat oppose, neither support or oppose, somewhat support, or strongly support. These responses were coded from 1 for strongly oppose, to 5 for strongly support. On average, respondents are largely neutral to redistribution with an average score of 3.3 out of 5. However, this masks a great deal of heterogeneity in the data. As Figure 1 shows, preferences for redistribution are actually bimodal, with 33 percent of respondents strongly supporting such redistribution and 23 percent of respondents strongly opposed to it. HETEROGENEITY IN ATTITUDES TOWARDS REDISTRIBUTION Standard utility maximization theory predicts the desire to redistribute income is strongly correlated with the respondent s own self-interest. Respondents who directly benefit from redistribution are likely to support it, while those who are harmed by redistribution are likely to oppose it. In the context of the health care proposal those earning $280,000 or more will lose money on net, while those with lower earnings are likely to gain from it. If pecuniary interests dominate decisions regarding redistribution, then all high-income earners should oppose redistribution and the aforementioned protests by Democrat Congressmen are justified. However, studies by Fong (2001), Alesina and La Ferrara (2005), Reed-Arthurs and Sheffrin (2010), and others have shown that direct pecuniary consequences are one of many factors influencing demand for income redistribution. Race, gender, views on the determinants of wealth, expected future income, political affiliation and knowledge about the tax system also play roles in the demand for redistribution. Are these other factors enough to overcome self-interest or does support for redistribution break down purely along income tax brackets? Figure 2 plots the level of support for redistribution by the rich (those earning $150,000 to $200,000 and those earning over $200,000) and the poor (those earning less than $15,000 and those earning $15,000 to $25,000). We see that while the majority of respondents take a self-interested position, about 25 percent vote contrary to direct pecuniary incentives. Of these, 15 percent of the poor strongly oppose redistribution, while 19 percent of the rich strongly support redistribution. At least in the abstract, not all high earners oppose redistribution through higher tax rates. Are the respondents who vote counter to pecuniary incentives fundamentally different from their more self-interested colleagues? Columns 1 and 2 of Table 1 compare the characteristics of high-income individuals who support redistribution with high-income individuals who oppose it. Two major differences emerge. First, those supporting redistribution are much more likely Figure 1: for Government Redistribution Note: 119 respondents who were not at all sure are excluded from the table. 166

3 Figure 2: for Redistribution by Income Group Income < $15, Income: $150,000 - $200, Income: $15,001 - $25,000 Income > $200,

4 NATIONAL TAX ASSOCIATION PROCEEDINGS Table 1 Respodent Characteristics by Income and for Redistribution Income Greater than Income Less that Strongly $150,000 Strongly Strongly $25,000 Strongly Demographics Age (Years) Age_18_29 8% 7% 5 1 Age_30_ % 11% Age_50_64 41% 47% 22% Age_65_Plus 6% 8% 12% Any Children 23% 12% 21% 8% 1 Number of Children Married 88% 72% 28% Single 6% 12% 36% 5 White 81% 8 73% 79% Education (Years) Employment Employed Full Time 57% 58% 18% 26% Retired 7% 13% 18% Looking For Work 8% 2% 14% 13% Income $213,725 $200,668 $14,799 $16,212 Political Affiliation Liberal (Scale 1 to 7) Democrat (Scale 1 to 7) Tax Practices Itemize Deductions 86% 96% 29% 12% 1 Standard Deduction 14% 4% 44% 8 2 Other Tax System Somewhat or Very Complex 99% 10 91% 88% Should Completely Overhaul Tax System 8 34% 2 54% 34% 1 Should Eliminate Estate Tax % 71% 1 Maximum Fair Tax Rate Value of All Gov t Benefits $17,279 $22,863 $4,527 $3,563 Personal Tax Rate Too High 94% 14% 2 44% 37% Decrease Services and Taxes 73% 6% 2 69% 2 Increase Services and Taxes 1% 58% 2 22% 2 Observation Difference in means test significant at the level. 2 Difference in means test significant at the 1% level. 3 Self-rated on a 7 point scale where 1 is extremely conservative and 7 is extremely liberal. to be affiliated with the Democratic Party and to have a self-described liberal ideology. ers average a self-declared 5.83 on a 7-point scale of liberal ideology, while the average member of the opposition scored herself as a This result is expected since redistributive policies are one of the largest differences between party platforms. Perhaps more interesting is the general satisfaction that those who support redistribution have with the tax system. Those opposed to redistribution are significantly more likely to feel the tax system needs a complete overhaul (85 percent compared to 34 percent), that their current tax rate is too high (94 percent compared to 14 percent) and the estate tax should be eliminated (100 percent compared to 25 percent). Those who support redistribution feel the maximum fair tax rate is 13 percentage points higher than those who oppose redistribution. Seventy-three percent of those opposed to redistribution feel taxes and benefits should be reduced 168

5 compared to 6 percent of those who support redistribution. Columns 3 and 4 compare low-income supporters and opponents of redistribution. We find supporters and the opposition differ in several demographic characteristics. Respondents opposed to redistribution are on average seven years younger than those in favor of redistribution. Young respondents tend to have a steeper earnings trajectory than those with more work experience. This difference in age is consistent with respondents considering future earnings as well as present income. Poor families with children are less likely to support redistribution. This is unexpected since these same poor families are the most likely to receive substantial net flows from redistribution through the Earned Income Tax Credit, Medicaid, and similar programs. Aside from demographics, there are other differences among low-income households. Political ideology and party affiliation are again strongly linked with the desire to redistribute wealth: those supporting redistribution are significantly more liberal and more likely to align with the Democratic Party. We again find views on tax rates and government provided social services to be significantly different at the 5 percent level. Of the poor who do not support redistribution, 69 percent think the government should reduce both taxes and services, while only 20 percent of those in favor of redistribution feel taxes and services should be reduced. Finally, a larger percentage of low-income earners opposed to redistribution itemize income tax deductions compared to those who support redistribution. There may be a systematic difference in employment or other circumstances that both cause people to itemize their income tax deductions and to oppose redistribution. For example, those who are self-employed may be more likely to itemize tax deductions. They may also be less risk averse and thus less likely to support social safety net programs and redistribution to the poor. BENEFITS AND REDISTRIBUTION Next we turn to whether respondents valuations of benefits received from government programs are correlated with support for either the current tax system or demand for additional redistribution. Little research has been done to understand how people value government services. Some government services, like national defense, are rarely purchased in a free market, and the value of others, like environmental regulation, varies substantially from person to person. There has been a long tradition of work that attempts to quantitatively estimate and apportion the value of government benefits, often along with taxes. 5 One common approach is to allocate the value of public goods in accordance with measures of use (e.g., road expenditures by number of miles driven a year) or in proportion to income (particularly for items like national defense, international affairs, and technology development.) Actual perceptions of benefits and apportioned values will differ if there are variations in valuation beyond family income or measures of use. For example, two individuals earning the same income and living in the same town may place a different value on local parks, public schools, and police protection. This heterogeneity may affect the individual s satisfaction with the tax system used to pay for these benefits and support for additional redistribution through government programs. The Tax Foundation 2009 survey provides a unique platform to answer questions about perceived benefits. Subjects were asked: Thinking about all the government services you use during a year national defense, roads, public schools, food stamps, Social Security, police protection and so on how much would you say these are worth to you in dollars for one year? That is, how much would you be willing to pay for all the services provided to you by governments for one year? The median respondent valued all government services at $2,000 and 50 percent of responses fell between $500 and $7,500. While this question offers unique data on benefit valuation, a few caveats must be discussed. First, respondents don t know exactly how much they receive in government benefits. They must estimate the value in the course of the survey, introducing measurement error. Additional measurement error is introduced by respondents propensity to round answers to the nearest $1,000. These errors will attenuate any estimated effects of perceived government benefits on either satisfaction with the tax system or with desired redistribution, two topics which we now explore. There are several paths through which perceived benefits may be related to satisfaction with the current tax system or with a desire for additional redistribution. Each of these paths has different implications for the relationships between perceived net benefits, satisfaction with the tax system and a desire for additional redistribution. 169

6 NATIONAL TAX ASSOCIATION PROCEEDINGS First, under the benefit view of taxation, taxes represent the price one pays for government services and, in equilibrium, taxes paid should be approximately equal to the value of government benefits one receives. According to psychological equity theory, when individuals feel they are unjustly rewarded or penalized relative to their peers, they will take action to restore the equity in their interpersonal relationships. Together these theories imply that individuals will seek parity between the taxes they pay and the benefits they receive. Deviations in either direction, as measured by the absolute value of benefits minus taxes, should lead to dissatisfaction with the current system and a desire for corrective action through changes in redistribution. In the context of the Tax Foundation 2009 survey, this would imply that poor people who pay more in taxes than they receive in benefits as well as rich people who receive more in benefits than they pay in taxes should support redistribution through a tax on high-income earners, since it will decrease the difference between benefits received and taxes paid. Second, if individuals take a purely self-interested benefit maximizing stance, they are more likely to support the current tax system if they receive more in benefits then they pay in taxes. The greater the net transfer they receive from government programs the more likely they are to support the status quo. In this approach, there is no apparent reason that a measure of net benefits should be related to demand for additional redistribution after controlling for income. Third, variation in non-measured characteristics of an individual may drive the value she places on benefits, her satisfaction with the current tax system, and her demand for additional redistribution. For example, a risk-averse individual will place a high value on government provided social safety net programs and is more likely to support the current tax system that provides these benefits. This respondent is also more likely to support additional redistribution through similar government programs. Under this channel, one would expect satisfaction with the current tax system and demand for redistribution to be positively correlated with the perceived value of government benefits. In practice, survey respondents may be influenced by all of these factors and many others. Table 2 contains the results of probit regressions on whether the respondent believes the tax system should be overhauled or the respondent somewhat or strongly supports for redistribution through a tax on high-income earners on variables including measures of benefits and net benefits 6 with a comprehensive set of controls, including income and demographics. We find that dissatisfaction with the current tax system is decreasing with the log of perceived benefits from government programs (column 1) and support for redistribution is increasing with the log of perceived benefits (column 4). These results are consistent with the third pathway that individuals who place a high value on government programs (beyond what is captured by income and demographics) are more likely to support the status quo and social safety net programs. Next, consider the net benefits variables in columns 2 and 3 of Table 2. Dissatisfaction with the current tax system is decreasing in net benefits but is not statistically significantly related to the absolute value of net benefit. This suggests that respondents take a self-interested view (and not an equity view) when determining satisfaction with the tax system. In column 5, net benefits and the demand for redistribution are not significantly related. This is, again, consistent with a self-interested stance. The final columns of Table 2 test whether preferences for redistribution are consistent with equity theory. Column 6 includes an indicator variable equal to one if the respondent has an income less than $50,000 and perceives net negative benefits or if the respondent has an income greater than $150,000 and perceives a positive net benefit. Under the equity view, both groups should support increasing redistribution through a tax on high-income earners because it brings taxes paid and benefits received closer to parity. The results indicate that poor respondents with negative net benefits are (insignificantly) more likely to support redistribution, but on average, rich respondents with positive benefits are less likely to support additional redistribution. Column 7 contains a similar experiment run in the opposite direction. Specifically, poor respondents who pay more in taxes than they receive in government benefits and rich respondents who receive more in benefits than they pay in taxes should support additional redistribution to the poor. Again we find insignificant results and conflicting signs on the coefficients. Overall we find that, satisfaction with the current tax system and demand for redistribution through taxation are indeed correlated with an individual s valuation of the benefits received from government 170

7 Table 2 Role of Perceived Benefits in for Taxation Overhaul Tax System Redistribution Dependent Variable: (1) (2) (3) (4) (5) (6) (7) Ln(Benefits) *** 0.041*** (0.007) (0.009) Net Benefits (000s) * (0.001) Absolute Value Net Benefits (0.000) (0.000) (Net Benefit < 0) * (Income <$50,000) (0.131) (Net Benefit > 0) * (Income >$150,000) (0.360) (Net Benefit > 0) * (Income <$50,000) (0.133) (Net Benefit < 0) * (Income >$150,000) (0.347) Observations Probit regression, reporting marginal effects. Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 Model includes controls for age, race, gender, marital status, number of children, income, education, region of residence and employment status. Omitted categories are male, single, white, age 18 to 29, less than HS education, income < $25,000, live in West. Taxes liability could not be estimated for 308 individuals. As such, specifications involving Net Benefits have fewer observations. programs. This relationship is consistent with underlying factors like risk aversion that influence all three variables simultaneously and with respondents seeking to maximize financial gain. We find, however, that the effects of an increase in benefits on the demand for redistribution are small relative to other factors like gender or political affiliation. ACTUAL VS. PERCEIVED GOVERNMENT BENEFITS There is virtually no research on how perceived benefits match actual benefits received by families or how close the standard methods for allocating benefits according to family income match peoples own perceptions. Given the available data, we conduct an initial analysis of the correlates of perceived benefits. If respondents were rational and had full knowledge, perceived benefits would be the sum of the quantity of each benefit received multiplied by the individual s valuation for that particular benefit. Demographic characteristics should also influence perceived benefits. Higher-income families may perceive more benefits from public goods such as education or the role of government in providing a stable framework to earn income. Families with children would likely place a higher value on public schools and parks. Poor families with children are more likely to be eligible for EITC and welfare programs, increasing the dollar value of benefits received. The unemployed are more likely to receive unemployment insurance and job training programs, but this effect may be mitigated if the unemployed are more likely to live in blighted areas which provide fewer public services or feel that the system has let them down. People over the age of 65 are more likely to be eligible for Social Security. Table 3 presents results of our exploration into public perceptions of benefits, reporting results of a regression of the log of perceived benefits on demographics. There are a number of striking findings. Higher-income individuals appear to place 171

8 Table 3 Correlates with Perceived Benefits Dependent Variable: Ln (Benefi ts) Income: Under $25, (0.250) Income: $25,000 - $50, (0.243) Income: $50,000 - $75, (0.265) Income: $75,000 - $100, ** (0.294) Income: $100,000 - $150, *** (0.289) Income: $150,000 - $200, *** (0.327) Income: Over $200, *** (0.359) Female *** (0.128) Married 0.318** (0.141) Black *** (0.290) Children 0.536** (0.227) Income Under $25,000 * Children (0.591) Age: 50 to *** (0.164) Age: Over *** (0.168) Age < 65 * Retired 0.404* (0.237) Looking for Work ** (0.252) Some College (0.163) College 0.788*** (0.203) Graduate Education 1.400*** (0.156) Democrat (Scale 1 to 7) (0.042) Liberal (Scale 1 to 7) (0.047) Observations 2001 R-squared Robust standard errors in parentheses, *** p<0.01, ** p<0.05. Omitted categories: income not reported, male, single, no children, not black, age 18 to 49, high school education or less.

9 a higher value on government benefits. The log of perceived benefits is strongly and consistently increasing in income in a near linear fashion. Men, married couples, and families with children all perceive higher levels of benefits. Respondents over 65 and those who have retired early also perceive higher levels of benefits than the non-retired and non-elderly. Black and unemployed respondents perceive lower levels of benefits than white respondents or employed respondents. This may be due to different areas of residence, or, alternatively a sense of disenfranchisement. On the other hand, simple political affiliation and ideology do not appear to be related to perceived levels of benefits. CONCLUSIONS Research on public attitudes does illuminate some aspects of the debate on taxes on high-income individuals to fund health care. We find there is substantial variation in levels of support within income groups and many wealthy Democrats support high-income taxes on high earners in the abstract. Thus, Democratic congressmen should hear a wide range of views even from their wealthy constituents. Taxpayers who perceive generally higher benefits from government programs are more likely to support redistribution, even after controlling for income. We estimate these effects to be small relative to factors like political ideology; however it may be understated due to measurement error. Finally, preliminary estimates suggest that perceived benefits from government programs appear to be reasonably related to objective factors determining benefits, but there are important deviations that require further exploration. Notes 1 For the quote and a discussion of this issue, see Weisman (2009, p. A1). As we shall see, even in abstract settings, there is considerable diversity among highincome individuals on the desirability of redistribution. 2 Maher (2009). In 2009, less than half of those were in California. 3 See Reed-Arthurs and Sheffrin (2010) and Hite and Roberts (1991). 4 Robert Reischauer expressed the first view, Leonard Burman, the second. See Calmes (2009). 5 For a survey of prior work, as well as some estimates, see Chamberlin and Prante (2007). For one recent application, see Prante and Fleenor (2009). 6 Net benefits are calculated as benefits less estimated total taxes, where total taxes have been estimated using the National Bureau of Economic Research TaxSim program. References Alesina, Alberto and Eliana La Ferrara. Preferences for Redistribution in the Land of Opportunities. Journal of Public Economics 89 (June 2005): Bartlett, Bruce. Tax the Rich. Forbes, July 10, Calmes, Jackie. Obama s Pledge to Tax Only the Rich Can t Pay for Everything, Analysts Say. New York Times, August 1, Chamberlin, Andrew and Gerald Prante. Who Pays Taxes and Who Receives Government Spending? An Analysis of Federal, State, and Local Tax and Spending Distributions Washington, D.C.: Tax Foundation, Tax Foundation Working Paper 1. Fong, Christina. Social Preferences, Self-interest and the Demand for Redistribution. Journal of Public Economics 82 (November 2001): Hite, Peggy A. and Michael L. Roberts. An Experimental Investigation of Taxpayer Judgments on Rate Structure in the Individual Income Tax. Journal of the American Taxation Association 13 (Fall 1991): Maher, Lucy. America s Most Expensive ZIP Codes. Forbes, August 27, Moon, Matt. How Do Americans Feel About Taxes Today? Tax Foundation s 2009 Survey of U.S. Attitudes on Taxes, Government Spending and Wealth Distribution. Special Report No Washington, D.C.: Tax Foundation, Prante, Gerald and Patrick Fleenor. How Much Does President Obama s Budget Redistribute Income? Washington, D.C.: Tax Foundation, Tax Foundation Special Report 172. Reed-Arthurs, Rebbecca and Steven M. Sheffrin. Understanding the Public s Attitudes Towards Redistribution Through Taxation. New Orleans, LA: Tulane University, Economics Working Paper Weisman, Jonathan. Democrat s New Worry: Their Own Rich Voters. Wall Street Journal, July 20,